Pixel-level Scene Understanding in One Token: Visual States Need What-is-Where Composition

Add code
Mar 14, 2026

Share this with someone who'll enjoy it:

View paper onarxiv icon

Share this with someone who'll enjoy it: